one of the basic experimental designs that can be used to remove from the experimental error two sources of variation identified with the rows and columns of the square —note 1. this design should be used only if all interaction terms can be assumed to be zero —note 2. a 5 × 5 Latin square consists of 25 plots, which need not themselves be square, arranged in five rows of five plots to the row, in such a way that each of the five treatments occurs once in each row and columnThis definition last updated 07/12/2008